Mathematical Models of Perception and Generation of Art Works by Dynamic Motions

Abstract

This paper presents a study on the role of dynamic motions in the creation and perception processes of action-art paintings. Although there is a lot of interest and some qualitative knowledge around, there are no quantitative models in the scientific computing sense about this process yet. To create such models and implement them on a robotic platform is the objective of our work. Therefore, we performed motion capture experimentswith an artist and reconstructed the recorded motions by fitting the data to a rigid-body model of the artist’s arm. A second model of a 6-DOF robotic platform is used to generate new motions by means of optimization and optimal control algorithms. Additionally, we present an image analysis framework that computes certain image characteristics related to aesthetic perception and a web tool that we developed to perform online sorting and cluster studies with participants.We present first results concerning motion reconstruction and perception studies and give an outlook to what will be the next steps towards an autonomous painting robotic platform.